On First Partial Transcript, MAI-Transcribe-2-Streaming achieves a 2.5% WER at 0.12s after end of speech, ahead of Grok Voice Transcribe 2.0 at 3.4% WER and 0.49s, and ahead of Muse Voice Transcribe and ElevenLabs Scribe v2 Realtime on accuracy, both at 3.6%, and slightly faster than both at 0.12s versus 0.13s. It is also more accurate than Cartesia Ink-2 with external endpoints at 4.0% WER and 0.07s, and AssemblyAI U3.5 Realtime Pro, whose Min Latency and Max Accuracy modes both achieve 4.0% WER. It is the lowest-error point on the First Partial Pareto frontier.